1. Introduction

The Fishery Regulation Assessment Model (FRAM) is an accounting tool for estimating the impacts of proposed fisheries on Chinook (Oncorhynchus tshawytscha) and Coho (Oncorhynchus kisutch) salmon stocks during a fishing season. FRAM is a software program developed and used by:

  • the Pacific Fishery Management Council (PFMC)
  • the Washington Department of Fish and Wildlife (WDFW)
  • Washington Treaty Indian Tribes and the Northwest Indian Fisheries Commission (NWIFC)
  • the Pacific Salmon Commission (Coho only) (PSC)
  • and the U.S. National Marine Fisheries Service (NMFS)

FRAM is a single-season modeling tool with separate code for Chinook and Coho salmon. The Chinook version evaluates impacts on most stock groups originating from the California Central Valley (Sacramento River), north-central Oregon coast, Columbia River, Willapa Bay, north Washington coast, Puget Sound, and southern British Columbia. The Coho version evaluates impacts on a comprehensive set of stocks originating from Central California to Southeast Alaska and is considered to represent total Pacific West Coast production. FRAM results are used to examine the impacts of proposed fisheries within a single management year for compliance with management objectives, allocation arrangements, Endangered Species Act (ESA) requirements, and domestic and international legal obligations.

This site includes FRAM model descriptions of FRAM data, program flow and algorithms, a user manual focused on running the FRAM application, a description of the FRAM base period (under construction), and information about pre- and post-season use of the model.

2. Background

FRAM is a deterministic, discrete, time-age structured model that aids in the planning of Chinook and Coho salmon fisheries. It is designed to address the effects of multiple mixed-stock fisheries on individual stocks, each further stratified by age and mark status (and possibly hatchery origin). Individual stock-age groups are exploited as a single pool, that is, in each time step all pre-terminal fisheries operate on the entire cohort simultaneously. The model calculates fishery related mortality, including catches and non-landed impacts, imposed on the abundance of each stock unit.

In a given yearly model run, over the course of several time steps, FRAM allocates a mixed stock cohort to catches, non-retention mortality and escapement. This occurs by combining stock-specific run sizes and fishery inputs with parameters calibrated from observations made during a fixed “base period” (set of baseline years). The base period integrates fishery catches, stock age and mark composition information, and coded wire tag (CWT) recovery data to produce a calibrated reference pattern of stock distributions and stock-specific harvest rates by time-step and fishery.

A fundamental element of FRAM calculations are the stock-age-fishery-timestep specific exploitation rates (defined as the total mortality divided by abundance). This expression of catch as a proportion of the available abundance is based primarily on coded wire tag release and recovery data from the Regional Mark Information System (RMIS).

3. History

FRAM is an empirically fit tool that uses the past to inform a representation of the future. Over time, algorithms have been added or adapted to capture changing management options, and the development platform of the application has evolved, but the basic structure remains much the same as when it was conceived in the late 1970’s.

In the mid-1970s, treaty fishery rights litigation, and the associated legal obligation for the states of Washington and Oregon to provide Treaty Tribes with the opportunity to harvest specific shares of individual runs, introduced a need for a quantitative tool to project the impact of proposed salmon fisheries at the stock-specific level. Other legal issues, such as the Magnuson-Stevens Fishery Conservation and Management Act and the Law of the Seas convention, contributed to the need to improve assessment tools. These considerations provided the impetus to develop early salmon fishery assessment models using the information available from the coast wide CWT program.

In the late 1970s, the Washington Department of Fisheries (WDF) and U.S. National Bureau of Standards (NBS) developed a model for evaluating alternative fishery regulatory packages that was coded in FORTRAN and ran on a mainframe computer at the University of Washington. The WDF/NBS Model could be configured for either Chinook or Coho by using different input data files. This WDF/NBS model was not extensively used by the PFMC because it proved costly to operate and results were difficult to obtain in a timely manner. Morishima and Henry (2000) provide a more in-depth history of Pacific Northwest salmon management and fishery modeling.

In the early 1980s, the development of personal computers allowed for the WDF/NBS model to be converted into simple spreadsheet models. This transformation improved accessibility to the model during the PFMC preseason planning processes. The first spreadsheet model for Chinook used by the PFMC was developed in the mid-1980s to model Columbia River “tule” fall Chinook. The corresponding spreadsheet model for Coho, the Coho Assessment Model (CAM), covered stocks from the Columbia River, Puget Sound, and Washington and Oregon coastal areas. The CAM was revised over time, principally to improve report generation capabilities and provide more detailed information for managing terminal area fisheries in Puget Sound using Terminal Area Management Modules (TAMMs). The CAM was used as the primary model for evaluating Coho impacts for proposed PFMC fisheries until the mid-1990s.

The increased complexity of proposed fishery regulation regimes, along with the need for increased temporal and spatial resolution for impact projections, soon surpassed the capability of the spreadsheet models. In the mid-1990s, CAM was programmed in QUICK BASIC and renamed FRAM. The recognition that common algorithms underlie both the Coho and Chinook spreadsheet models led to the effort to develop the QUICK BASIC version of FRAM for both species. The FRAM code could be used to evaluate proposed fishery regulation regimes for either Chinook or Coho by using different input file configurations. In 1998, FRAM was converted to VISUAL BASIC to take advantage of the improved user interface available through the Microsoft Windows operating system. A multi-agency Model Evaluation Subgroup periodically reviewed model performance and parameter estimation methods and coordinated revisions to the model during this period (1998-2000).

A new version of FRAM was introduced in 2012. It was coded by James Packer (WDFW) in VISUAL STUDIO.NET and uses a Microsoft Access database rather than binary text files for data storage and retrieval. In 2017, additional software changes were implemented when an updated Chinook base period was adopted using brood year 2005-2008 CWTs. Substantial changes were made to the calibration programs, modeling of sublegal encounters, growth estimation, imputing CWT recoveries for stocks and fisheries with few or no CWTs, handling mark-selective fisheries, and developing model stock proportions.

Although some program subroutines are shared, Chinook and Coho are processed separately, with distinct time step divisions, stocks, and fisheries. The Chinook version evaluates impacts on most stock groups originating from the California Central Valley (Sacramento River), north-central Oregon coast, Columbia River, Washington coast, Puget Sound, and Southern British Columbia. However, not all stocks along the West Coast of North America are represented in Chinook FRAM. Some of the stocks that are not represented in Chinook FRAM include any stock in California except Sacramento Falls, Oregon stocks originating south of Elk River, Mid and Upriver Columbia River Springs, Snake River Springs, and Canadian stocks originating north of Georgia Strait. The Coho version evaluates impacts on a comprehensive set of stocks originating from Central California to Southeast Alaska and is considered to represent total Pacific West Coast production.

4. FRAM Interactions with Other Models

Figure 1. Simplified interactions between the Base Period, FRAM, and TAMM (TRS: Terminal Run Size, BPER: Base Period Exploitation Rates)

4.1. Terminal Area Management Module (TAMM)

TAMM is a species-specific Microsoft Excel spreadsheet that interfaces with the FRAM application, providing additional input values and receiving output from FRAM following a run. The major purpose of TAMM/FRAM interactions is turning TAMM catches or harvest rates from terminal Puget Sound or Washington Coastal fisheries into stock-specific mortalities during a FRAM run. TAMM catches are simply reported or estimated landed catches from terminal fisheries regardless of stock origin. FRAM allocates impacts to stocks using TAMM fishery definitions. Once FRAM is run with TAMM, landed catches from TAMM fisheries are stored as quotas in the FRAM database and can be recalled during subsequent FRAM runs without the use of a TAMM file. While FRAM can run indepently of a TAMM file, almost all modeling exercises require a TAMM file. For a list of TAMM fisheries that are processed in FRAM see Appendix 3 and 4. For more information about TAMM see “Model Detail” chapter 5 for Coho and Chinook.

4.2. FRAM Base Peiod

The FRAM base period is developed by summarizing coded-wire-tag information from a continuous time period that needs to contain a sufficient number of CWT releases and recoveries. The current Chinook base period uses CWT releases from 2005-2008. The Coho base period is comprised of CWT recoveries from catch years 1986-1992. A calibration program maps the CWTs to FRAM fisheries, stocks, and time steps and conducts a cohort reconstruction to estimate base period exploitation rates and other parameters, such as base period cohort sizes. These estimates form the basis of a FRAM run and remain static between model runs. Using these base period parameters allows FRAM to predict the stock composition, exploitation rates, and escapements of future fisheries when provided with forecasts of abundance and fishery mortalities.

5. Coho vs Chinook Differences

While there is only one FRAM program that handles both Chinook and Coho model runs and many of the core processes and algorithms are identical, this document presents the model and computational structures independently for each species. This decision was made in an effort to improve readability and clarity, as there are a variety of unique features specific to each species that can be difficult to differentiate when they are presented together.

Table 1. Differences between Chinook and Coho FRAM

Description Chinook Coho
Number Stocks 78 246
Number Fisheries 73 198
Time Steps 4 5
Ages Age 2-5 Age 3
Maturation Yes No
Sublegals Yes No
Adult Equivalents Yes No
Size Limit Modeling Yes No
Growth Modeling Yes No
Time Period October (yr-1) - April (yr+1) January - December
Mark Selective Fishery Bias Correction No Yes
BkFRAM Starts with terminal run size Starts with escapement
Stocks Major stocks affecting PFMC and Washington fisheries All stocks
Fisheries Major marine fisheries affecting PFMC and Washington stocks All fisheries
Areas Marine Marine and Freshwater
Model Stock Proportion (accounting for non-model stocks) Yes No
Post-fishing Abundance reported to TAMM Marine escapement Escapement
Forecast Units Mainly terminal run size Ocean age-3 (majority)
Flexible Terminal Run Size Definitions No Yes (TAAETRSList table)


To cite this page:
Salmon modeling and analysis workgroup. 2023. FRAM Overview in FRAM Documentation. https://framverse.github.io/fram_doc/ built September 21, 2023.


Previous Documentation

FRAM Overview 2008